706 Parallel Coordinates

library(plotly)
library(dplyr)

#plotly code:
plotlyOutput('pcp')

output$pcp <- renderPlotly({
        
 
      pcdat<- dat[1:54,c(2,18,27, 29)] %>% subset(!is.na(dat[1:54,29]))
      party <-dat[1:54,29] %>% subset(!is.na(dat[,29])) %>%  unique() 
      col_names <- sapply(pcdat, is.factor)
      pcdat[,col_names] <- sapply(pcdat[,col_names] , unclass)
      
      library(plotly) 
      pcdat %>% plot_ly() %>% add_trace(type = 'parcoords', line = list(color = ~Governor.Political.Affiliation, colorscale = list(c(0,'blue'),c(1,'red'))),
                                        dimensions = list(
                                          list(range = c(1,50),
                                               tickvals = c(1:50),
                                               label = 'state',
                                               ticktext = c(paste(dat$state)),
                                               values = ~state ),
                                          
                                          list(range = c(~min(total),~max(total)),
                                               label = 'total tests',
                                               tickvals = c(~min(total),~max(total)),
                                               values= ~total),
                                          
                                          list(range=c(1,2),
                                               tickvals = c(1, 2),
                                               label = 'party',
                                               ticktext = c(paste(party)),
                                               values = ~Governor.Political.Affiliation ),
                                          
                                          list(range = c(1,7),
                                               tickvals = c(1,7),
                                               label = 'school closure date',
                                               ticktext = c(paste(dat$StateClosureStartDate)),
                                               values = ~StateClosureStartDate
                                          )
                                        )
                                        
                                        
      )
      
    })

  fig <- pcdat %>% plot_ly(width = 1000, height = 600) 
      #fig %>% 
      pcdat<- dat[1:54,c(2,18,27, 29)] %>% subset(!is.na(dat[1:54,29]))
      party <-dat[1:54,29] %>% subset(!is.na(dat[,29])) %>%  unique() 
      col_names <- sapply(pcdat, is.factor)
      pcdat[,col_names] <- sapply(pcdat[,col_names] , unclass)
      
      library(plotly) 
      #options(viewer=NULL)
      pcdat %>% plot_ly() %>% add_trace(type = 'parcoords', line = list(color = ~Governor.Political.Affiliation, colorscale = list(c(0,'blue'),c(1,'red'))),
                                        dimensions = list(
                                          list(range = c(1,50),
                                               tickvals = c(1:50),
                                               label = 'state',
                                               ticktext = c(paste(dat$state)),
                                               values = ~state ),
                                          
                                          list(range = c(~min(total),~max(total)),
                                               label = 'total tests',
                                               tickvals = c(~min(total),~max(total)),
                                               values= ~total),
                                          
                                          list(range=c(1,2),
                                               tickvals = c(1, 2),
                                               label = 'party',
                                               ticktext = c(paste(party)),
                                               values = ~Governor.Political.Affiliation ),
                                          
                                          list(range = c(1,7),
                                               tickvals = c(1,7),
                                               label = 'school closure date',
                                               ticktext = c(paste(dat$StateClosureStartDate)),
                                               values = ~StateClosureStartDate
                                          )
                                        )
                                        
                                        
      )

NA

fig <- pcdat %>% plot_ly(width = 1000, height = 600) 
#fig %>% 
pcdat<- dat[1:54,c(2,18,27, 29)] %>% subset(!is.na(dat[1:54,29]))
party <-dat[1:54,29] %>% subset(!is.na(dat[,29])) %>%  unique() 
col_names <- sapply(pcdat, is.factor)
pcdat[,col_names] <- sapply(pcdat[,col_names] , unclass)

library(plotly) 
#options(viewer=NULL)
pcdat %>% plot_ly() %>% add_trace(type = 'parcoords', line = list(color = ~Governor.Political.Affiliation, colorscale = list(c(0,'blue'),c(1,'red'))),
dimensions = list(
  list(range = c(1,50),
       tickvals = c(1:50),
       label = 'state',
       ticktext = c(paste(dat$state)),
       values = ~state ),
  
  list(range = c(~min(total),~max(total)),
       label = 'total tests',
       tickvals = c(~min(total),~max(total)),
       values= ~total),

list(range=c(1,2),
     tickvals = c(1, 2),
     label = 'party',
     ticktext = c(paste(party)),
     values = ~Governor.Political.Affiliation ),

list(range = c(1,7),
     tickvals = c(1,7),
     label = 'school closure date',
     ticktext = c(paste(dat$StateClosureStartDate)),
     values = ~StateClosureStartDate
                  )
     )
                  
                
)

An example parallel coordinate plot to use as a template. This code is from Plotly documentation: https://plotly.com/r/parallel-coordinates-plot/

df <- read.csv("https://raw.githubusercontent.com/bcdunbar/datasets/master/parcoords_data.csv")

fig <- df %>%
  plot_ly(width = 1000, height = 600) 
fig <- fig %>% add_trace(type = 'parcoords',
          # line = list(color = ~colorVal,
          #             colorscale = 'Jet',
          #             showscale = TRUE,
          #             reversescale = TRUE,
          #             cmin = -4000,
          #             cmax = -100),
          dimensions = list(
            list(range = c(~min(blockHeight),~max(blockHeight)),
                 constraintrange = c(100000,150000),
                 label = 'Block Height', values = ~blockHeight),
            list(range = c(~min(blockWidth),~max(blockWidth)),
                 label = 'Block Width', values = ~blockWidth),
            list(tickvals = c(0,0.5,1,2,3),
                 ticktext = c('A','AB','B','Y','Z'),
                 label = 'Cyclinder Material', values = ~cycMaterial),
            list(range = c(-1,4),
                 tickvals = c(0,1,2,3),
                 label = 'Block Material', values = ~blockMaterial),
            list(range = c(~min(totalWeight),~max(totalWeight)),
                 visible = TRUE,
                 label = 'Total Weight', values = ~totalWeight),
            list(range = c(~min(assemblyPW),~max(assemblyPW)),
                 label = 'Assembly Penalty Weight', values = ~assemblyPW),
            list(range = c(~min(HstW),~max(HstW)),
                 label = 'Height st Width', values = ~HstW),
            list(range = c(~min(minHW),~max(minHW)),
                 label = 'Min Height Width', values = ~minHW),
            list(range = c(~min(minWD),~max(minWD)),
                 label = 'Min Width Diameter', values = ~minWD),
            list(range = c(~min(rfBlock),~max(rfBlock)),
                 label = 'RF Block', values = ~rfBlock)
            )
          )


fig

head(df)

On to our code. Merge the two datasets and join them on the state abbreviation.

Define the regions based on CDC standards:

#define regions based on CDC standards (https://www.cdc.gov/coronavirus/2019-ncov/covid-data/covidview/04102020/labs-regions.html)

R1 <- c('CT', 'ME', 'MA', 'NH', 'RI', 'VT')
R2 <- c('NJ', 'NY', 'PR')
R3 <- c('DE', 'DC', 'MD', 'PA', 'VA', 'WV')
R4 <- c('AL', 'FL', 'GA', 'KY', 'MS', 'NC', 'SC', 'TN')
R5 <- c('IL', 'IN', 'MI','MN','OH','WI')
R6 <- c('AR','LA','NM','OK','TX')
R7 <- c('IA', 'KS', 'MO', 'NE')
R8 <- c('CO', 'MT', 'ND','SD','UT','WY')
R9 <- c('AZ', 'CA', 'GU', 'HI','NV')
R10 <- c('AK','ID','OR','WA')

#put as character so that all states/territories have a value and all values of same type of vector
dat <- dat %>% mutate(region = (
  case_when(state %in% R1 ~ '1',
            state %in% R2 ~ '2',
            state %in% R3 ~ '3',
            state %in% R4 ~ '4',
            state %in% R5 ~ '5',
            state %in% R6 ~ '6',
            state %in% R7 ~ '7',
            state %in% R8 ~ '8',
            state %in% R9 ~ '9',
            state %in% R10 ~ '10', 
            TRUE ~ 'Other')
))

#Reference Documentation : https://plotly.com/r/reference/#parcoords

#DO NOT RUN YET. have not updated completely
fig <- dat %>%
  plot_ly(width = 1000, height = 600) 
fig <- fig %>% add_trace(type = 'parcoords', name = ~state,
          dimensions = list(
            list(tickvals = c(0:50),
                 ticktext = c(dat$state),
                 label = 'State', values = ~state),
            list(
              range=c(~min(totalTestResults), ~max(totalTestResults)),
              label = 'Total Tests',
              values = ~totalTestResults)
            
            # list(range = c(~min(blockHeight),~max(blockHeight)),
            #      constraintrange = c(100000,150000),
            #      label = 'Block Height', values = ~blockHeight),
            # list(range = c(~min(blockWidth),~max(blockWidth)),
            #      label = 'Block Width', values = ~blockWidth),
            # list(tickvals = c(0,0.5,1,2,3),
            #      ticktext = c('A','AB','B','Y','Z'),
            #      label = 'Cyclinder Material', values = ~cycMaterial),
            # list(range = c(-1,4),
            #      tickvals = c(0,1,2,3),
            #      label = 'Block Material', values = ~blockMaterial),
            # list(range = c(~min(totalWeight),~max(totalWeight)),
            #      visible = TRUE,
            #      label = 'Total Weight', values = ~totalWeight),
            # list(range = c(~min(assemblyPW),~max(assemblyPW)),
            #      label = 'Assembly Penalty Weight', values = ~assemblyPW),
            # list(range = c(~min(HstW),~max(HstW)),
            #      label = 'Height st Width', values = ~HstW),
            # list(range = c(~min(minHW),~max(minHW)),
            #      label = 'Min Height Width', values = ~minHW),
            # list(range = c(~min(minWD),~max(minWD)),
            #      label = 'Min Width Diameter', values = ~minWD),
            # list(range = c(~min(rfBlock),~max(rfBlock)),
            #      label = 'RF Block', values = ~rfBlock)
            )
          )


fig

GGally more difficult to do categorical data, but way to separate out groups, scale the data (univariately, substract mean & divide by sd)

library(GGally)
dat %>% arrange(desc(region))%>%
   ggparcoord(
    columns = c(18,29), groupColumn = 32
    ) +
    scale_color_manual(values=c( "#69b3a2", "#E8E8E8", "#E8E8E8", "#E8E8E8", "#E8E8E8", "#E8E8E8", "#E8E8E8", "#E8E8E8", "#E8E8E8", "#E8E8E8", "#E8E8E8") ) 

 #can factor regions
#dat %>% colnames()

THIS ONE!! cdparcoord